Method for Matrix Data Broadcast in Parallel Processing

2. The apparatus as recited in claim 1, wherein the circuitry is configured to generate the single request responsive to a qualifying condition being satisfied.

6. The apparatus as recited in claim 3, wherein the threshold number of threads is stored in a field.

8. The apparatus as recited in claim 7, wherein the cache controller is configured to convey the mask during the broadcast of the given data to the compute units.

10. The method as recited in claim 9, further comprising generating the single request responsive to a qualifying condition being satisfied.

11. The method as recited in claim 10, wherein the qualifying condition is the plurality of threads including at least a threshold number of threads.

13. The method as recited in claim 11, further comprising preventing sending an access request targeting the given data, responsive to determining the plurality of threads include fewer than the threshold number of threads.

16. The computing system as recited in claim 15, wherein the instructions are executable by the processing unit to generate the single request responsive to a qualifying condition being satisfied.

17. The computing system as recited in claim 16, wherein the qualifying condition is the plurality of threads includes at least a threshold number of threads.

19. The computing system as recited in claim 18, wherein the single request comprises the broadcast identifier.